[
Lists Home |
Date Index |
Thread Index
]
Jens:
First let me thank you for starting this thread. Since I have very
little poetic talents, I was feeling hopelessly inadequate watching the
display of talent on the "Thread of the Week", the Limerick thread.
Even though your post is bit of a flame bait, I feel I at least have
something to say on this.
Several points you make are correct:
1. XML is definitely not new.
It can be argued that CSV or other Flat files are XML with invisible
tags that are already agreed upon by the receiver and the sender.
2. There is nothing that has not been doable for eons with RPC and ASCII
Yes but that was a world without the web and without inter-enterprise
computing. Here I would like to paraphrase something Jon Bosak once
told me. Think about using RPC as the primary means of transactions
between 2 business partners. Now imagine that due to some problem in
the code, the transactions are now in dispute and the aggrieved parties
take the case to court for remedy. How reasonable is it to expect the
legal system to dig through the court, find the problem and presribe a
remedy to this? On the other hand if this was done in the loosely
coupled sense using XML messages over any transport and the request and
responses are logged, it is certainly more reasonable to expect the
court system to have an easier time figuring out the issues and
prescribe a remedy. So when it gets to inter-enterprise and between
smoke stacks or silos within an enterprise, XML may be a better approach
as opposed to RPC.
I believe:
1. When you are dealing with disparate applications, data sources over
which you have no control and you have to serve applications and devices
over which you have no control, XML will certainly ease a lot of your
burden. As sure as the sun coming up every day, these sources,
applications and devices will change and will change often. The cost to
you if you use XML as the plumbing is going to be lot less.
2. When you are dealing with a situation where you are the "master of
your own destiny" (borrowed this from Mike Champion) and do not foresee
this changing soon, stick to time tested methods of SQL, RPC and ASCII.
You will do just fine.
Now think of a situation where all applications, databases and
repositories miraculously allow read and write in XML (I can not think
of why that would be even preferable), all of a sudden the "switching
cost" that keeps you locked into a particular vendor or application for
good disappears. The IP of the spreadsheet, data or the document you
create is yours. Then tell me why you have to pay vendors money to be
able to manipulate it as you wish.
Use XML as little or as much as you want. Your requirements,
environment and infrastructure and above all ROI should dictate how deep
you want to go into this. I for one believe that I should dip my toes
in the water a little, test it and check the benefits and then if things
check out, dip my foot and so on.
Finally to give you some practical perspective, we at B-Bop, in some
cases, have helped solve customer problem with nothing more than a
parser, a transformation tool and a few XML files. While at others, we
have used XPath, our homegrown XML Query Language and the full power of
our XML Data Management System to solve a different problem.
So while the peaks and valleys of any new technology can be
disconcerting and disappointing at times, enjoy the ride but always make
sure you are getting your money's worth. If I were a betting man, I
would bet that this ride is far from being over. It is only getting
started, doubts and concerns notwithstanding.
Soumitra
|
